[Condor-users] Negotiator dies repeatedly on a subset of machines

["Alan Cass" <alan.ocais@xxxxxxxxx>]

Hi all,

On a subset of machines in our Windows pool the negotiator dies repeatedly with errors like "The NEGOTIATOR (pid 4140) died due to exception ACCESS_VIOLATION" in the MasterLog. Because of this the machines accept a job but immediately go into preemption and stick in that state for hours on end. Any idea what might cause such behaviour to suddenly start happening? 

Alan